tailored algorithms — Aj Collins Artistry/Pexels

Australians could soon be able to choose what appears in their social media feed. How would this work?

Author Daniel Angus & Ashwin Nagappa
Date 1 September

On ABC’s Insiders program on Sunday, federal Social Services Minister Tanya Plibersek said the government is exploring whether Australians should be able to opt in to algorithmically curated social media feeds.

The idea has been championed by sexual consent activist Chanel Contos. Her organisation, Teach Us Consent, is running a campaign which calls for personalised recommendation algorithms to become opt-in. Instead, users would have, by default, a chronological feed containing posts from accounts they have decided to follow.

There are some positive aspects to the proposal. But a chronological feed is not really a “non-algorithmic” feed. It still relies on an algorithm, albeit an exceptionally simple one: take a collection of posts, order them according to time, and show the newest first.

This might sound pedantic. But it points to something missing from the current debate. The problem is not that social media platforms use algorithms. Almost every digital system we interact with uses algorithms to decide what information to sort, filter, or display.

The better question is what are we asking those algorithms to optimise for, and who gets to decide.

How we got to this point

Early social media feeds were relatively simple. You followed accounts and their posts appeared, generally in reverse chronological order.

Over time, and largely driven by commercial imperatives, platforms moved towards increasingly sophisticated and secretive recommendation algorithms. Today, much of what appears in our social media feeds may come from accounts we have never encountered before.

Contemporary recommendation algorithms use our individual and collective data to make predictions about what will keep us clicking, sharing and scrolling. That has produced some genuinely useful forms of discovery. But maximising engagement is a poor substitute for maximising social value.

Indeed, some of the things most effective at capturing our attention are those that make us angry, outraged or frightened. Sensational claims, interpersonal conflict and material deliberately constructed to provoke a response can all perform remarkably well in an attention economy.

Concerns become even more serious when recommendation algorithms repeatedly introduce people to misogynistic, extremist or conspiratorial content.

Making personalised recommendations opt-in could potentially be a worthwhile intervention, although the devil will be in the detail of exactly what ranking mechanism platforms would be required to provide instead. At the very least, it restores an important degree of agency. People should have meaningful choices about the systems organising their attention.

But chronological feeds aren’t a silver bullet.

Chronological doesn’t necessarily mean good

Before platforms became so heavily driven by automated recommendation, discovery was often a much more social process. We all had a few friends who reliably found the interesting, funny, strange or important things happening online and passed them along.

In a sense, we were the algorithms, collectively filtering and curating the internet for one another through our social networks.
There is real value in recovering some of that. It makes the feed more legible because we have some understanding of why something has appeared in front of us.

The earliest social bookmarking services on the web, like Delicious and Digg, let us curate and share interesting content, independent of the platforms or source website.

But chronology alone cannot recreate that environment.

A feed is the end point of a much larger information system involving search, advertising, trending topics, influencers, reposting, moderation and, importantly, the kinds of content people encounter elsewhere and then bring back into their networks.

Even a strictly chronological feed can contain rage bait, misinformation and misogyny if those materials are already circulating through the accounts somebody follows. That distinction matters for regulation.

What if we could tailor our feed?

Social media users in some jurisdictions already have some control over their feeds. For example, the European Union’s Digital Services Act requires large social media platforms to give users the option of selecting a feed not based on algorithmic profiling.

Some of the most interesting experiments are happening outside the dominant platforms.

Mastodon has generally emphasised chronological timelines containing posts from accounts users have deliberately followed. But this creates its own challenges. For example, arriving on a new network without knowing who to follow can make the place feel remarkably empty. At the other extreme, Mastodon discovered through their own study how users found it difficult to keep up with the volume of their feed as they followed more accounts.

Bluesky has taken a different approach. Alongside a chronological following feed, users can select from thousands of custom feeds created by other people and organisations.

This suggests another possibility beyond choosing between “the algorithm” and “no algorithm”: algorithmic choice.

Rather than Meta, TikTok or X determining the single feed optimisation system through which millions of people experience their platforms, users might choose between different ways of organising information.

A news organisation might maintain a feed prioritising verified reporting. A university could create one highlighting research. Communities could build feeds around particular shared interests. Users could then choose a simple chronological feed, or indeed several different feeds for different purposes.

Bluesky has described this as an “open marketplace of feeds”. Mastodon is experimenting with complementary approaches to discovery and community curation. Neither provides a finished regulatory blueprint. But together the wider decentralised social media ecosystem is becoming a useful live experiment in what happens when we separate the social network from the mechanism used to rank its content.

This is where the Australian debate could become much more interesting.
Giving people the ability to turn off personalised recommendation systems would be a sensible start. But we should also be asking why a handful of technology companies have such extraordinary control over the systems that organise our attention in the first place.

Instead of treating chronological ordering as the end point, we could look more closely at actively encouraging alternative open infrastructure that is resistant to future corporate capture and embodies genuine choice for community.

Daniel Angus, Professor of Digital Communication, Director of QUT Digital Media Research Centre, Queensland University of Technology and Ashwin Nagappa, Postdoctoral Research Fellow, ARC Centre of Excellence for Automated Decision-Making and Society, Queensland University of Technology

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Return to news.